We are studying mechanisms by which mRNAs are selected for degradation by the nonsense-mediated mRNA decay (NMD) pathway. One major clue towards an understanding of this problem is the observation that a unifying characteristic of mRNAs degraded by NMD is the presence of a long 3' untranslated region (3'UTR). We have previously presented evidence that a core component of the NMD pathway, the RNA helicase Upf1, is able to act as a direct sensor of 3UTR length. Upf1 appears to accomplish this function by interacting sequence-nonspecifically with RNA, leading to preferential accumulation on long 3UTRs. We are using retroviral RNA elements that are able to inhibit decay of long 3'UTR-containing transcripts as tools to understand mechanisms of substrate selection by Upf1. These elements block the process of decay at distinct steps in the pathway, providing opportunities to probe protein and RNA requirements for both the initial steps of 3'UTR length sensing and the transition between Upf1 binding to mRNA and the initiation of decay. First, we are using an 400 nt segment of the Rous sarcoma virus (RSV) gag-pol mRNA, termed the RNA stability element (RSE), that has previously been shown to protect viral RNAs from Upf1-dependent decay. We have found that the RSE can inhibit decay of well characterized reporter mRNAs containing long 3'UTRs in human cells and are investigating its impact on the recruitment of Upf1 to mRNAs. In addition, we have identified proteins that interact with and mediate the function of the RSE. Using RNAseq to characterize the gene expression profiles of cells in the presence or abscence of RSE-binding factors has revealed that the mechanism employed by RSV to protect its RNAs from decay is shared with as many as several hundred human mRNAs. We are currently working to understand the biochemical mechanisms and cellular consequences of this novel mode of gene expression regulation. Second, we have reported that a translational readthrough-promoting RNA pseudoknot from the Moloney murine leukemia virus inhibits NMD. Translational readthrough events induced by the pseudoknot appear to inhibit the process of NMD at two distinct steps. Under conditions of frequent readthrough, steady-state Upf1 accumulation in mRNPs is reduced, impairing 3'UTR length sensing. Interestingly, when the pseudoknot is used to stimulate low levels of readthrough (i.e. 1%), Upf1 accumulation in mRNPs is restored, but decay remains inhibited. We have now developed a system that allows concurrent assessment of RNA stability, translational readthrough, and Upf1 binding in order to better understand the relationship between translational readthrough and decay inhibition. We are using this system to probe the effects of readthrough stimulated via distinct mechanisms and to biochemically identify functional intermediates in the NMD pathway.